Joe Benton

Karma: 101

Won’t vs. Can’t: Sandbagging-like Behavior from Claude Models

Joe Benton and Zachary Witten

19 Feb 2025 20:47 UTC

15 points

1 comment1 min readLW link

(alignment.anthropic.com)

Sabotage Evaluations for Frontier Models

David Duvenaud, Joe Benton, Sam Bowman, evhub, mishajw, Eric Christiansen, HoldenKarnofsky, Ethan Perez and Buck

18 Oct 2024 22:33 UTC

94 points

56 comments6 min readLW link

(assets.anthropic.com)